The stack behind agent-made video.

Command-line downloader. A separate MCP server exists for metadata and subtitles. The star count is the yt-dlp repo itself. Unlicense. Download only what you have rights to.

Node graph for diffusion image and video workflows. GPL-3.0.

The usual WebGL runtime when an agent builds a 3D scene in the browser. MIT.

Open-source timeline editor, MIT. 24fps says an MCP server and headless batch render are on the roadmap, not shipped, so no agent is marked supported yet.

Grant Sanderson's original Manim, MIT. The community edition is the one under active release tags.

HTML video framework built for coding agents, with an official skill pack and a component registry.

Agent video production system with many pipelines and skill files. AGPL-3.0.

React framework for programmatic video. Source is public under the Remotion License, not an OSI open-source license. Free for individuals and organizations of up to 3 people; larger for-profits need a company license.

Community edition of the math animation engine. MIT.

Hugging Face library that runs many open video pipelines. Apache-2.0.

Community MCP plugin so an agent can build and animate scenes in Blender. MIT. Not made by the Blender Foundation.

Original Lottie player for the web. MIT.

Agent skill that transcribes, cuts filler, grades, captions, and renders a folder of footage with FFmpeg. MIT.

Animation library used inside agent-made motion graphics. The core repo has no SPDX id. The official agent skills repo is MIT.

Open-Sora training and inference code. Apache-2.0.

Open-source text-to-speech with voice cloning, MIT, used for voice-over in agent videos.

Stability's generative-models repo, which ships Stable Video Diffusion. MIT.

CTranslate2 implementation of Whisper for local transcripts. MIT.

Speech-to-text with word timestamps, the usual source for caption timing in agent edits. BSD-2-Clause.

Creative-coding library. 24fps indexes clips made with it. LGPL-2.1.

TypeScript motion-graphics library. The latest GitHub release is v3.17.2 from December 2024.

Portrait animation that transfers expression onto a still. Public repo, no SPDX id from the GitHub API.

Alibaba Wan 2.2 weights. Apache-2.0. Distinct from the Wan 3.0 API prices above.

Alibaba Wan 2.1 weights. Apache-2.0. This is the public repo, not the Wan 3.0 API rate.

FramePack sampling for long local video generations. Apache-2.0. Latest GitHub release tag was 'windows'.

Shot-recipe cards and Remotion implementations for product videos.

Tencent's public Hunyuan3D 2 repository. No SPDX id from the GitHub API.

Python library for cuts, composites, and captions. MIT. A common target when an agent writes an edit script.

macOS editor with a local MCP server so an agent can read the timeline. GPL-3.0.

Microsoft's image-to-3D model. MIT on GitHub.

Audio-driven talking head. Public repo, no SPDX id from the GitHub API.

Lip-sync a face video to a new audio track. Public repo, no SPDX license from the GitHub API.

Zhipu's CogVideo family. The GitHub repo is Apache-2.0.

Tencent's public HunyuanVideo repository. GitHub did not return an SPDX license, so it is not marked open source.

AnimateDiff motion module for image models. Apache-2.0.

A community runner for Wan-family models on smaller GPUs. No SPDX id from the GitHub API.

Lightricks LTX-Video weights and code. The GitHub license is Apache-2.0.

Lightricks LTX-2 repository. GitHub returned no SPDX id on 2026-11-03, so open source is left false.

Short films drawn in code across many style prompts, with directing and sound notes.

Audio-driven portrait animation. MIT.

SkyReels V2 public repository. No SPDX license came back from the GitHub API.

Real-time lipsync. Public repo, no SPDX id from the GitHub API.

Latent lipsync model. Apache-2.0.

Animates in a large set of art styles and rejects films where almost nothing moves.

Command-line cutter that drops silent stretches and can export an editor timeline. Unlicense.

HTML and CSS to MP4 for coding agents. Apache-2.0. 24fps describes it as shipping templates.

Open-source programmatic video framework, MIT, forked from the Motion Canvas line.

Mermaid-to-SVG renderer used for diagram scenes. Public repo, no SPDX id from the GitHub API. 24fps describes it as closed source.

Alibaba's VACE all-in-one video creation and editing repo. Apache-2.0.

Community MCP server for DaVinci Resolve Studio timelines, markers, and renders. MIT. Not made by Blackmagic.

Genmo's Mochi 1 open-weight video model, Apache-2.0 on GitHub.

Motion films that stay in one continuous camera move instead of cutting between slides.

Desktop editor where the timeline is code, with MCP, skills, and a CLI. Engine and editor repos are MPL-2.0.

Gives an agent ears: finds moments in a recording, measures loudness, and critiques a soundtrack. CPU only. Star count is from the 24fps video-tools page, which attributes stars to the GitHub API. This snapshot did not re-query that repo.

Rust video framework on a Skia renderer, MIT, with an agent skill.

Motion principles for agents: timing, easing, and choreography.

Manim-style Python API on a Rust and GPU core, MIT. 24fps lists it as an agent-oriented renderer.

Video production toolkit skill for Claude Code, including ffmpeg and caption steps.

Editorial halftone paper-collage b-roll with approval gates before it animates.

Official MCP server for voice-over, sound effects, transcription, and voice design. MIT.

Procedural single-file canvas videos for Claude Code, with style and storyboard check-ins.

A taste pass for agent-made motion graphics.

dotLottie web player. MIT. 24fps indexes Lottie clips as a motion library.

Vchitect 2.0 public code. Apache-2.0.

Community MCP that drives After Effects compositions through its scripting API. Not made by Adobe. Star count is from the 24fps video-tools page, which attributes stars to the GitHub API. This snapshot did not re-query that repo.

Community MCP for editing in Premiere Pro. 24fps lists Claude and Codex. Not made by Adobe. Star count is from the 24fps video-tools page, which attributes stars to the GitHub API. This snapshot did not re-query that repo.

Motion design toolkit covering review loops, sound, and web simulations.

Motion design in HTML, Playwright, and ffmpeg, without After Effects or a Remotion license.

Skill that walks Claude through Remotion motion graphics, captions, and sound.

Runs Manim scenes from an agent and returns the rendered video. Star count is from the 24fps video-tools page, which attributes stars to the GitHub API. This snapshot did not re-query that repo.

MCP server for building and changing TouchDesigner networks. Star count is from the 24fps video-tools page, which attributes stars to the GitHub API. This snapshot did not re-query that repo.

Remotion components for product demos, installed as source through the shadcn CLI, with agent skills. MIT.

Skills for launch films, charts, and milestone reveals, rendered as code.

Code-first video rendered on WebGPU, with compositor skills.

Video engine that accepts Remotion-style projects. 24fps reports a PolyForm Small Business license, and GitHub returned no SPDX id, so it is not marked open source.

The agent writes a plan file that renders through HyperFrames, FFmpeg, and whisper.cpp. MIT.

Local studio: a one-sentence brief in, a rendered video out, without API keys. MIT.

Open-source editor whose timeline can be driven from an MCP client. MIT.

Uploads, searches, and edits video through Video Jungle. Closed source on the 24fps listing. Star count is from the 24fps video-tools page, which attributes stars to the GitHub API. This snapshot did not re-query that repo.

A field guide of motion techniques, GSAP snippets, and timed recipes for agents.

Local MCP server for FFmpeg edits, subtitles, audio, and HyperFrames compositions. Apache-2.0.

SRT-driven b-roll that routes each shot to HyperFrames or Remotion.

Skill for Motion.dev web animation: springs, scroll, and gestures.

SKILL.md for creating Manim animations. Also the install command on the Manim Community row.

Terminal video tool for silence cuts, captions, and storyboards, with a cost cap on your own API keys. MIT.

Controls OBS scenes, sources, and recording from an agent. Star count is from the 24fps video-tools page, which attributes stars to the GitHub API. This snapshot did not re-query that repo.

Khan Academy style explainer: a pen draws while a voice explains. Local, no API keys.

Beat-synced HyperFrames motion graphics with voice-over, captions, and music.

Motion graphics skill with a gallery of looks for Claude Code.

HTML explainer boards synced to a Whisper transcript and exported with Chrome and FFmpeg.

Local text-to-speech with the Kokoro model for voice-over drafts. Star count is from the 24fps video-tools page, which attributes stars to the GitHub API. This snapshot did not re-query that repo.

Deterministic video rendering with Vue.

Motion components for Remotion, delivered as source. MIT.

Script, storyboard, and render narrated explainers and diagram films.

Open caption format: one word stored once, then cut for 16:9 or 9:16. MIT. Ships an agent skill.

Skill that points DeepSeek at Remotion motion prompts.

Finds public-domain b-roll from the Internet Archive and writes a provenance record next to the clip. Star count is from the 24fps video-tools page, which attributes stars to the GitHub API. This snapshot did not re-query that repo.

Remotion motion graphics skill aimed at cleaner agent output.

Beat-synced type for social, titles, and lyric videos.

Apple-style kinetic typography recaps in Remotion.

Launch films, explainers, and data stories as HTML and MP4.

Procedural animation in Rust, Apache-2.0. Early, and the API still moves between versions.

Small Odin and Lua renderer that turns a video script into an MP4. MIT.

Kuaishou's Turbo Pro video model, billed per second on Runway Dev. The pricing table does not state the output resolution.

Kling 3.0 on Runway Dev, with separate per-second rates when audio is off and on. Resolution is not stated on that table.

Kling 3.0's 4K tier on Runway Dev. The page says the rate does not change when audio is toggled.

Kling O3 on Runway Dev. Standard, Pro, and 4K are separate rates, and video-to-video is billed on the input duration.

Kling 4.0 on Runway Dev, with audio always on. The standing list rate is 15.7 credits/s at 720p and 23.5 at 1080p. A separate 20% promo through 30 November 2026 UTC is not the standing price.

Alibaba Wan 3.0 as priced on Runway Dev, with separate 480p, 720p, and 1080p rates.

Wan 3 Prime on Runway Dev. Higher per-second rates than Wan 3.0 at the same labeled resolutions.

ByteDance Seedance 2.5 on Runway Dev. Output seconds are billed by resolution, and input or reference video adds a second charge.

Seedance 2.0 on Runway Dev, with published 480p, 720p, 1080p, and 4K per-second rates and a 56 credit minimum.

Faster Seedance 2.0 tier on Runway Dev. The table lists 480p and 720p, not 1080p or 4K.

Seedance 2.0 Mini on Runway Dev, priced at 480p and 720p with a 28 credit minimum.

xAI's Grok Imagine Video 1.5 on Runway Dev, with 480p, 720p, and 1080p rates plus one credit per image or audio reference.

The lite Grok Imagine Video tier on Runway Dev. 720p is 3 credits per second before the start-frame credit.

MiniMax Hailuo 3 on Runway Dev. Rates are labeled 768P and 2K, so they are not filed under 720p, 1080p, or 4K.

MiniMax H3 Max on Runway Dev, priced at 480p and 768p. Those labels are not 720p, 1080p, or 4K.

Alibaba's HappyHorse video model. Runway Dev prices 720p and 1080p. Alibaba's API reference lists duration and aspect ratio, including 9:16.

Google Veo 3.1 on Runway Dev, with a higher rate when audio is on. The table does not attach those rates to a resolution.

The fast Veo 3.1 tier on Runway Dev. Audio and no-audio rates are both published, without a resolution label.

Gemini Omni Flash video on Runway Dev. Text-to-video and image-to-video are 10 credits per second, plus a first-frame credit for image-to-video.

Gemini Omni Flash 1.1 on Runway Dev, with separate 360p, 720p, 1080p, and 4K rates plus reference charges.

Runway's Gen-4.5 video model on the developer API at 12 credits per second. The pricing row does not name a resolution.

Runway Gen-4 Turbo on the developer API at 5 credits per second. Resolution is not on that row.

Runway Aleph 2.0, priced at 28 credits per second with a 56 credit minimum. Used for video editing and restyling rather than a fresh text-to-video clip.

Runway Act-Two performance model at 5 credits per second on the developer price table.

OpenAI's Sora 2. The public pricing page checked on 2026-11-03 did not show a video per-second rate, so price and current API status are left blank.

Pika's video product and effects tools. No per-second price was taken from an official page in this pass.

Luma's Ray video models. The docs page checked on 2026-11-03 pointed at billing but did not expose a per-second dollar rate in the extract.

PixVerse text-to-video and effects product. Price fields are blank because an official per-second table was not captured in this pass.

Shengshu's Vidu video model. Listed without a price because no official per-second figure was captured on 2026-11-03.

Hosted inference for video, image, and audio models, with a per-model price on each model page. Those model pages did not yield a clean rate table in this pass, so prices are blank.

Hosted model API. The pricing page fetched on 2026-11-03 did not expose per-second video rates in the extract, so prices are blank.

Runway's developer API. Credit price used for the model rows is $0.01 per credit, from the pricing guide.

Aggregator for video generation models. No per-second rate was copied because the vendor table was not parsed in this pass.

Inference host that also serves some media models. Video per-second prices were not verified here.

Inference platform. No video per-second price was verified for this listing.

Inference host. Video prices left blank pending a parsed official table.

Model hub and hosted inference. The diffusers library is listed separately.

Serverless GPU platform used to host custom video pipelines. Not a per-second model price.

Model API host. Prices not copied in this pass.

Media model API. Prices not copied in this pass.

Aggregator across video and image vendors. Prices not copied in this pass.

GPU cloud and model API. Video per-second prices not verified here.

Inference platform used for open video models. Prices not verified here.

Alibaba's API for Wan and HappyHorse. Dollar prices on this listing come from Runway's resale table, not from a Model Studio price sheet.

Google's API for Gemini and Veo. The developer pricing page was fetched, but the Veo dollar rows were not isolated, so this host has no price fields.

MiniMax API for Hailuo video and speech. Per-second video prices on model rows are Runway resale rates.

Kuaishou's Kling product and API. Per-second figures on the Kling model rows are Runway Dev rates.

Open-source 3D suite. Agents drive it through the community Blender MCP, listed separately. License is GPL; star count is not taken from the Blender repo in this snapshot.

Cloud render API for programmatic edits, with a CLI and an agent skill. The editor product is closed source.

Finds a motion graphic, prompt, or editor preset in the 24fps library before the agent writes one.

Avatar and agent video platform. HyperFrames, the open-source framework, is a separate listing. API dollar rates were not on the pricing page captured 2026-11-03.

Enterprise avatar video platform. No per-second price was captured.

Character and talking-head generation. Price left blank.

Talking-head and photo-animation API. Price left blank.

Replica and conversational video API. Price left blank.

Avatar training and workplace video. Price left blank.

Avatar ads from a script. Price left blank.

Transcript-based editor for cuts, filler words, and studio sound. Price left blank.

Mobile editor for captions, clips, and talking avatars. Price left blank.

Turns long videos into short clips. Price left blank.

Clipping and captions for long-form video. Price left blank.

Animated captions and short-form edits. Price left blank.

Clip finder for podcasts and YouTube. Price left blank.

Browser editor with captions, translate, and cleanup tools. Price left blank.

Browser editor for resize, captions, and smart cuts. Price left blank.

Editor and template library used for short-form video. Price left blank.

Prompt-to-video editor with a stock and template library. Price left blank.

Script to video with voice-over. Price left blank.

Blog and zoom-recording to video, with captions. Price left blank.

Speech, voice cloning, and sound effects. The API page shows character and minute pricing, which is not a video per-second rate, so those fields are blank.

Low-latency speech API used for real-time voice. Per-second video price does not apply. Speech price was not copied.

Music generation. Not priced per video second.

Music generation. Not priced per video second.

Stability's audio and music model. Video per-second fields left blank.

Speech and voice cloning API. Price left blank.

Text-to-speech API. Price left blank.

Text and image to 3D meshes. Not a per-second video price.

Image and text to 3D. Price left blank.

Hyper3D Rodin image-to-3D. Price left blank.

Browser 3D design tool with interactive embeds. Price left blank.

Upscale, denoise, and frame interpolation for footage. Price left blank.

Template-based video render API. Price left blank.

JSON timeline to MP4 API. Price left blank.

After Effects project templates rendered through an API. Price left blank.

Interactive state-machine animation, exported into apps and video. Price left blank.

Browser motion-design tool for interfaces and social posts. Price left blank.

Marketplace and tools for Lottie motion. The motion-design skill is a separate listing.

Image and video template API for generated marketing assets. Price left blank.

Camera-move and video generation product. An agent skill repo is listed on 24fps. Price left blank.

Stylized video generation. Price left blank.

Character animation that transfers motion onto a figure. Price left blank.

Lipsync API. Price left blank.

Face swap, avatar, and video translation tools. Price left blank.

AI-assisted character animation in 3D. Price left blank.

Library of AI-made motion graphics with prompts, plus an MCP server and a skill that searches the library before an agent writes a new film.